Improving Energy Efficiency in Iron Ore Sintering through Segregation: a Theoretical Investigation

Abstract

In iron ore sintering, effective segregation of the particulate bed on the strand can result in productivity increases and decreased fuel rate. To find the optimum level of segregation for a given blend, intensive experimental work will be necessary to assess the impact of different segregation levels. The impact of segregation on sintering performance can also be quantified using mathematical modelling. In this work, a well-validated sintering mathematical model has been developed and the effects of mean granule size, bed voidage, bed bulk density, coke mass segregation as well as increased bed permeability on the sintering performance have been investigated. It was concluded that the variation in coke mass down the bed and increased bed permeability are the major factors giving increased sinter yield and productivity in segregated beds. Coke mass segregation has a large impact on maximum bed temperature, residence time above the critical temperature and, consequently, the total heat available to the bed at the critical melt formation period. The effect of variations in granule size and bulk density, caused by segregation, on coke combustion efficiency and bed temperature is small. Bed permeability has the largest impact on flame front speed and, therefore, sinter productivity. Based on differences in bottom and top bed temperatures, the optimal coke mass segregation level was identified. Results from this study provide useful guidelines on optimal segregation level to maintain sintering performance and reduce energy consumption and therefore ironmaking costs.

1. Introduction

Iron ore sintering is an important pre-processing technology in integrated steel mills. Sinter accounts for around 70 wt.% of the ferrous feed to blast furnaces in the Asia-Pacific region.¹⁾ Sinter making is a complex process and can be viewed as comprising bed preparation under ambient conditions (or cold processing) and subjecting the bed to high temperatures by generating a narrow, moving flame front (or hot processing). The cold processing includes the blending of iron ore fines, fluxes, fuel and cold return fines, followed by the granulation process and charging of the formed granules onto the sinter strand. The hot processing starts with the establishment of a flame front on the top surface of the bed under an ignition hood followed by the progression of the front down through the bed, melt generation within the flame front and finally the solidification of melt. More detailed descriptions of the iron ore sintering process are available in the literature.^2,3,4)

It is well-known that the properties of the green granulated bed, which are dependent on the cold processing conditions, have a significant influence on flame front properties, heat generation, the amount of heat transferred to the solids and bed temperatures. For a given ore blend, granulation conditions can determine the size distribution of the granules and the location of coke particles – which has an influence on combustion rate and efficiency.^5,6) At a plant, the granulated sinter mix discharges from the feed hopper onto an inclined rill plate, and from there onto the moving strand. Granules segregate during this process because of differences in size and density. The level of segregation determines the heterogeneity of the formed bed and there are benefits in forming such a bed. For example, enhancing segregation will increase coke levels in the upper bed and decrease coke levels in the lower bed. This will help provide more even sintering temperatures which will help improve yield and sinter quality and ultimately improve the energy efficiency of the process. Assuming a typical fuel rate of 55 kg/tonne and a global sinter production of 1 billion tonne per annum leads to a consumption of 55 million tonne per annum (mtpa) of coke breeze, which corresponds to approximately 200 mtpa of CO₂ emissions to the atmosphere. Clearly, sintering is a significant energy consumer and greenhouse gas producer and improvements in energy efficiency need to be targeted.

With segregation, sinter yield from the upper bed increases as a result of increased heat generation from coke combustion. At the same time air temperatures are reduced in the lower bed and this prevents over-melting of the sinter mix and the formation of dense sinters, which do not reduce readily in a blast furnace. This has an additional positive impact on fuel rate in the blast furnace.

The aim of the present work is to further study the effect of bed properties changes caused by segregation (i.e., differences in granule size, coke mass, bed voidage and bulk density) on energy utilization and sintering performance. A well-validated hot model^1,5,7) has been formulated to describe heat transfer from the flame front to the bed and is used in this study. Data on segregated beds from the literature is used in the simulations. The impacts of the factors were considered individually and then combined to determine their cumulative effect on heat transfer during sintering.

2. Segregation in Iron Ore Sintering

2.1. Segregation Process

Iron ore sintering mills utilise a variety of segregation techniques to charge granules onto the strand. Mills segregate in order to improve bed permeability, as well as improve fuel efficiency. The effectiveness of the segregation process determines granule size distribution in the vertical direction down the bed. This essentially means that chemical composition, bed bulk density, bed voidage and, therefore, bed permeability^8,9,10,11) are not uniform down a green granulated bed. Overall, bed permeability improves because a narrower size distribution is obtained at every level down the bed and this leads to improved sinter plant productivity. For this reason both theoretical and experimental investigations of segregation in iron ore sintering have attracted attention since the 1950s.¹¹⁾ Many proprietary segregation devices have been introduced to promote the segregation of granules in a mix as it is being laid onto the sinter strand to form a bed. The optimum level of segregation will be dependent on the ore blend.

Figure 1 illustrates the size segregation process in sintering. On leaving the feed hopper, a drum feeder sends the granules rolling down an inclined rill plate (or charging chute) onto the moving strand. The drop height on leaving the rill plate is greater than the height of the bed, which can be close to 1 m. Larger particles have a greater momentum running down the sloping surface of the formed pile and as a result they travel a further distance down the strand before coming to a halt. As a result there is a greater concentration of larger granules in the bottom regions of a green granulated bed. It is possible to facilitate segregation by altering the angle of the rill plate but if it is too horizontal the flow of material will not be smooth and avalanching (continual material build up and collapse) will be an issue. Clearly the feed rate and drop height will also determine the plate angle required to optimise segregation. It is also important to remember that the process also depends on the properties of the ore blend since traded iron ores have very different size distributions and bulk densities.

Fig. 1.

Schematic diagram showing the discharge of granules onto the strand and the mechanism of segregation.

2.2. Literature Review

O’Dea and Waters^8,9) developed a simplified segregation model for iron ore sintering. Granule size distribution as a function of bed depth was determined and the predicted mean granule size down the bed agreed well with the experimental data. To simulate charging from a typical roll feeder, Nakano et al.¹²⁾ employed the discrete element method (DEM) to determine the size segregation in the vertical direction of packed bed. They concluded that size segregation increases with decreasing chute angle. DEM was also used by Ishihara et al.¹³⁾ to investigate the influence of bed collapse (avalanching) on size segregation. To date DEM studies have only been carried out to determine the effect of size on segregation. For a model to be useful in sintering it must also have the ability to predict the segregation of the different components in the sinter mix. Machida et al.¹⁰⁾ studied segregation of sinter mixes high in goethite using a pilot-scale sinter pot and also at plant level. They found that optimum coke segregation profile - defined in terms of maximum sinter shatter strength - varied with the sinter mix goethite level. Lower goethite content mixes gave better results when coke segregation increased. Plant trials on optimising coke segregation also showed improvements in tumble index and productivity. The density of goethite is less than that of hematite but many traded goethitic ores are coarse and this means increased goethite levels in the lower bed.

For a multi-component sinter mix, the size distribution and specific gravity of each material can be very different, and this can also influence segregation. O’Dea⁸⁾ found in their experiments that the levels of coke, CaO and total Fe content varied down a bed. Ball¹¹⁾ observed the segregation of solids - including iron ore, return fines, fluxes and coke – and also moisture. In both cases, it was observed that coke content decreased down the bed with segregation. This is because coke distribution followed the segregation pattern of granules below 1/8 inch (3.175 mm).¹¹⁾ In sintering, coke with a nominal size of minus 3 mm is used and their hydrophobic surfaces mean that the larger particles are not well adhered within a granule.²⁾ Coke also has a lower specific gravity of around unity compared to iron ores which have values of 3.5 to 4, and this means granules containing significant coke are lighter. Reducing granule density and mean size will reduce the velocity and momentum of the particles during charging (assuming that they move unhindered) and they do not end up as far down the strand (Fig. 1). As coke is the primary energy provider in a sintering bed and has a large influence on the process, there are significant benefits in controlling its distribution down a bed using appropriate segregation techniques.

With segregation, large granules tend to accumulate in the lower bed and smaller granules in the upper bed. As expected, there is a granule mean size gradient down a bed. Compared with non-segregated bed, tests involving freezing granules in liquid nitrogen before sizing showed that granule size distribution is narrower at each bed depth in segregation.^8,9) For this reason, localised bed voidage values and the averaged value for the entire bed increases.^14,15,16,17) As a result, segregation results in a more permeable green bed, and this has been observed in both laboratory and plant sintering operations.^8,9,11,18)

In summary, the size and density of granules determine their movement during charging and this causes the formation of segregated beds. These beds have the following characteristics:

(1) granule mean size increasing down the bed;

(2) bed voidage increasing at every bed level;

(3) bed bulk density decreasing at every bed level; and

(4) overall bed permeability increasing.

2.3. Definition of Segregation Level

It is difficult to characterise the level of segregation in a bed.¹¹⁾ Until this is possible the efficiency of different feeding devices and the level of segregation achieved in different plants cannot be compared. Clearly, the properties of ore blends will also determine how amenable they are to segregation. This means that for a given bed it is appropriate to define segregation level with respect to the same bed packed homogeneously. As indicated, many parameters - chemical analyses, or physical properties like granule size or bulk density - can be used to define segregation level. In this study they are termed segregation parameters. In most cases, segregation parameters would change at different rates as size segregation in a bed increases. For this study, an overall segregation level γ (e.g. 1, 2, 3, 4) is used to indicate the extent of segregation in the bed. Additionally, a specific segregation level σ expressed as a percentage is employed to define segregation in size, mass (or chemical composition) and bed voidage. For parameters that are not directly related to segregation (e.g., sintering airflow rate), discrete values are adopted. All segregation parameters would change depending on the overall segregation level γ.

From sinter pot and plant data, O’Dea et al.^8,9) derived descriptions of granule size, mass and bed voidage segregation, and sintering airflow rate. These segregation parameter values are used in the analysis. A moisture level of 5.5 wt.% is assumed and the corresponding gradient for granule Sauter Mean Diameter (SMD) is 2.16 mm/m (obtained at a high segregation level and defined⁸⁾). Voidage of the whole bed increased from 0.383 to 0.394 and green bed airflow rate increased by 12.9% as segregation increased. Tables 1 and 2 summarize the input simulation data for the non-segregated (Base case) and segregated cases, respectively. The data for γ = 1 and 4 are essentially based on the findings of O’Dea et al.^8,9) By assuming linear variations in the value of these parameters down the bed, the specific segregation levels for size σ_s and mass σ_m are defined as:   

$${\sigma }_s=\frac{{\mathrm{\Gamma }}_{s\text{,}\mathrm{\hspace{0.17em} }btm}-{\mathrm{\Gamma }}_{s\text{,}\mathrm{\hspace{0.17em} }top}}{\text{2}}$$(1)

  

$${\sigma }_m=\frac{{\mathrm{\Gamma }}_{m\text{,}\mathrm{\hspace{0.17em} }top}-{\mathrm{\Gamma }}_{m\text{,}\mathrm{\hspace{0.17em} }btm}}{\text{2}}$$(2)

where, Γ_s is the ratio of granule size of the segregated case to the non-segregated case, and Γ_m is the ratio of component mass of the segregated case to the non-segregated case. Subscripts s, m, btm and top indicate size, mass, bottom and top of the bed, respectively. The definitions are illustrated in Fig. 2, where H′ is the normalized distance down bed and SMD is used to characterise granule size. Mass segregation refers to the variation in mass of coke down the bed. Examples of how segregation is described in Table 2 and Fig. 2 relative to the Base case (Table 1) are illustrated below:

Table 1. Parameters of the non-segregated case (Base Case) used for simulations.

Parameter	Value	Unit
Bed height (incl. hearth layer)	570	mm
Hearth layer thickness	30	mm
Bed ignition temperature	1200	°C
Ignition suction	6	kPa
Sintering airflow rate (25°C, 16 kPa)	164	m3 h−1
Granule SMD	2.3	mm
Bed voidage	0.383	–
Granule apparent density	3396	kg m−3
Coke mass fraction in sinter mix	3.26	wt.%
Moisture mass fraction in sinter mix	5.5	wt.%

Table 2. Parameters of the segregated cases for simulations.

Parameter	Value
Overall segregation level γ (-)	1*	2	3	4
Granule size segregation level σs (%)	0	6	12	18
Bed voidage σv (%)	0	0.97	1.93	2.90
Coke mass segregation level σm (%)	0	7	14	21
Sintering airflow rate, (m3 h−1)	164	168	171	174

*  Base case (non-segregated bed)

Fig. 2.

Definitions of the specific segregation levels of granule size σ_s and coke mass σ_m.

Granule size segregation – for a γ value of 4, σ_s = 18% means the granule SMD at bed top is 18% smaller than the non-segregated case. At the bottom there is a corresponding 18% increase in SMD. The changes in granule size vary linearly down the bed as indicated in Fig. 2.

Voidage segregation – for a γ value of 3, σ_v = 1.93% means that the voidage of the bed is 1.93% higher than the base case value of 0.383. The voidage value is kept unchanged over the entire height of the bed because an explicit relationship between granule size distribution and bed voidage is not available.

Coke segregation – for a γ value of 4, σ_m = 21% means that there is 21 wt.% more coke at the top of the bed compared to the non-segregated case. At the bottom there is a corresponding 21 wt.% decrease in coke. The changes in coke levels vary linearly down the bed as indicated in Fig. 2.

Sintering airflow rate - the increased sintering airflow rate due to increased bed permeability was derived from the flame front resistance model by using the following equation:¹⁹⁾   

$$V_s=V_g-k_{\text{5}}{V}_g^3$$(3)

where V_s is the sintering airflow rate (m³ h⁻¹) at the normal suction of 16 kPa, V_g is the green bed airflow rate (m³ h⁻¹) at the same suction, and k₅ is a constant relating to flame front resistance, which is 3.2×10⁻⁵ (h² m⁻⁶) under the normal coke rate condition. At the overall segregation level γ of 4, based on an increase in green bed airflow rate of 12.9% (from γ of 1) the calculated sintering airflow rate is 174 m³ h⁻¹. The increase in green bed airflow rate is assumed linear and interpolated values are used to determine the sintering airflow rates for the other two overall segregation levels. Tables 3 and 4 show the granule and coke size distributions and the chemical composition of raw materials used in this study. The other inputs required to run the model are obtained from a recent sinter pot program.

Table 3. Granule and coke size distribution.

Granule size (mm)	Cumulative mass percent passing (%)	Coke size (mm)	Cumulative mass percent passing (%)
16	100	8	100
11.2	93.75	6.3	99.50
8	86.52	3.15	87.31
5.6	78.08	1.18	59.12
4	59.80	0.6	45.25
2.8	38.26	0.3	33.03
2	25.61	0.15	21.25
1.4	14.80	0.106	16.34
1	7.75	0.075	12.54
0.71	3.70	0.063	11.66
0.5	1.68
0.355	0.63
0.25	0.20
0.15	0.06

Table 4. Chemical composition of raw materials.

Material	Total Fe	SiO2	Al2O3	CaO	MgO
Ore A	65.39	3.65	0.91	0.04	0.03
Ore B	66.39	1.24	1.09	0.03	0.03
Limestone	0.40	0.80	0.30	54.39	0.46
Serpentine	5.80	42.19	2.24	2.05	34.10
Dolomite	0.54	1.04	0.34	30.31	20.60
Coke Breeze	0.57	6.43	3.49	0.33	0.09

3. Modelling Heat Treatment in Segregated Beds

The developed mathematical model of the iron ore sintering process^1,5,20) was used to quantify the effect of segregation on heat transfer and flame front properties. Modifications were added to take into account the variations in the segregation parameters down a bed and also variations in sintering airflow rate. The model has been well validated against more than twenty carefully-controlled sinter pot tests,²⁰⁾ and fairly good agreement has been achieved between model and experimental temperature profiles and waste gas compositions. The sinter pot test conditions used for model validation are similar to those used in this study. The raw materials were mixed and granulated under the same standard procedures. The granules were charged into a cylindrical sinter pot (600 mm in height and 330 mm in diameter) to form a green bed. The sinter pot was well-insulated around the wall to reduce heat losses. An annular layer of fine ore was placed between the green bed and the pot wall to reduce the ingress of air as bed shrinks. This method has been proved effective in improving the accuracy of airflow rate measurement.⁵⁾ For bed temperature measurement, the thermocouples were made using 0.5 mm diameter platinum and platinum/13% rhodium wires. The wires were protected in sheaths and inserted into three bed locations. Waste gas analysis was carried out using the Gasmet DX-4000 gas analyser with Fourier transform infrared (FTIR) spectroscopy. The same sintering procedure was followed in all tests. A flame front was formed on the surface of the green bed using an ignition hood combusting natural gas. A fan located at the gas exhaust pipe provided the suction required to draw the flame front through the bed. After burn-thorough, the sinter bed was left to cool in the flowing air stream before discharging for processing.

A schematic diagram of the sinter pot considered in this work is shown in Fig. 3. The cylindrical sinter pot has a height of 570 mm and diameter of 330 mm. The hearth layer consists of sinter particles larger than 5 mm and has the thickness of 30 mm. The sinter pot was equally divided into three sections (top, middle and bottom) along the bed height. Sintering performance at the centre of each of the three sections was simulated and then averaged to determine a mean value for the entire bed. In the same figure, the effect of bed temperatures caused by the arrival and departure of the flame front, as recorded by inserted thermocouples, is also shown. The important parameters characterising heat transfer to the bed at the critical melt formation period (>1100°C) are listed in Table 5. The maximum temperature (MT) recorded, the residence time (RT) above 1100°C and also the enclosed area (EA) provide information on the bed heat treatment process as the flame front passes the area. A comparison between typical experimental and model temperature profiles for a non-segregated bed is given in Fig. 4, which shows fairly good agreement.

Fig. 3.

Schematic diagram of the sinter pot used in this study and three temperature profile parameters, i.e., MT, RT and EA. Temperature profiles at three equally-spaced positions are recorded.

Table 5. Sintering performance indicators used in this study.

Parameter	Symbol	Definition
Maximum Temperature (°C)	MT	The maximum temperature in the temperature profile (see Fig. 3)
Residence Time (min)	RT	The residence time above 1100°C in the temperature profile (see Fig. 3)
Enclosed Area (min°C)	EA	The area enclosed by temperature profile and the y = 1100°C line in the temperature profile (see Fig. 3)
Flame Front Speed (mm min−1)	FFS	The travelling speed of flame front from start of sintering to the end, defined as bed height divided by sintering time
Combustion Efficiency, (-)	η	The volumetric concentration ratio of [CO2]/([CO2]+[CO]) produced from coke combustion

Fig. 4.

Comparison between typical experimental and model temperature profiles at top, middle and bottom of the bed.

In sintering, flame front speed (FFS) and coke combustion efficiency (η) have a large influence on the three flame front parameters shown in Fig. 3, and they are also defined in Table 5. Unless indicated otherwise, only averaged values of MT, RT, EA, FFS and η for the entire bed are reported in this study. The effects of segregation on MT, RT, EA, FFS and η are considered in this study. Initial discussions will explore the effect of varying the segregation parameters in isolation to understand their individual impact on the parameters in Table 5. Following the study, all the changes in the segregation parameters are integrated to obtain an overall impact.

4. Results and Discussion

4.1. Effect of Granule Size Segregation

The effect of granule size segregation on MT (maximum bed temperature) distribution down the bed is shown in Fig. 5. To eliminate the effects of bed ignition and hearth layer, only the maximum temperature between the top and bottom thermocouples (0.095 to 0.475 m down the bed) is shown. Assumed size segregation levels, σ_s, are shown in Table 2. In the simulations, all other parameters including bed voidage, coke mass and sintering airflow rate were kept unchanged from the base case (non-segregated).

Fig. 5.

Effect of granule size segregation on MT distribution between the top and bottom thermocouples down the bed.

It is seen in Fig. 5 that maximum bed temperature increases down the bed, which is to be expected because of the preheating of the air before the flame front. The figure also shows that increasing segregation to increase the mean size of granules down the bed raised the upper bed temperatures. Temperature increases are more pronounced in the upper bed than the lower bed. For this simulation airflow rate and voidage are kept unchanged so the increase is a direct result of more effective convective heat transfer. This term is inversely proportional to granule mean size. With segregation, the granule mean size decreases in the upper bed leading to an increase in heat transfer in the upper bed. Large decreases in lower bed temperature are not observed because of the increase in convective heat transferred to this region from the upper region. In sintering the upper bed is weak and a large proportion of material from here ends up as return sinter fines.²¹⁾ This means that increasing the upper bed temperature gives a great benefit even though temperature in the lower bed is slightly reduced. At a specific size segregation level, a minimum exists in the MT chart, which is below the upper surface of the bed. This region is shielded from radiation during ignition and, unlike lower regions in the bed, does not benefit much from the pre-heating effect.

Figure 6 shows the averaged values of convective heat load and flame front properties for the top, middle and bottom sections of the bed. Fig. 6(a) shows that with increased granule size segregation the convective heat load ahead and behind the flame front increases, resulting in higher upper bed temperatures (Fig. 5). Convective heat values are much higher (about an order of magnitude) ahead of the front because it is a measure of heat flowing to this region from the intense flame front. Behind the front, the convective heat load is that flowing from the cooling sinter to preheat the air for combustion in the flame front, and it is not surprising that this value is comparatively smaller. It is important to note that the heat load changes are not large - less than 10% behind the flame front and less than 5% ahead of the flame front. Nonetheless, the increase in convection heat loads behind the flame front raised flame front temperatures and also convection heat loads out of the flame front (Fig. 6(a)).

Fig. 6.

Effect of granule size segregation level σ_s on convection heat transfer ahead of and behind flame front, as well as the sintering performance indicators.

The other three figures in Fig. 6 show the effect of size segregation on flame front properties and coke combustion efficiency. Figure 6(b) shows that averaged maximum temperatures MT increased and this can be attributed to increased convective heat load. The high temperature has also increased the efficiency of the oxidation of CO generated on the coke surface to CO₂. Hence the efficiency of coke combustion η increases. It is also important to note that these changes in maximum temperature and combustion efficiency are not large. The increased convective heat load with increased segregation has resulted in slight increases in flame front speed and also a wider temperature profile (Figs. 6(c) and 6(d)). This is possible because the change in flame front speed is only small, just over 1% (Fig. 6(c)). The resulting small increase in RT together with higher MT values has meant slightly larger EA values (Fig. 6(d)). This is expected to positively impact sinter strength.

Another benefit of reducing granule size in the upper bed is that it promotes sintering reactions, melt generation, inter-particle bonding and coalescence because smaller particles have larger contact surface per unit volume than large particles.¹¹⁾ For this reason, sinter yield in the upper bed increases through size segregation.

4.2. Effect of Bed Voidage and Bulk Density

In practice, bed voidage actually changes with granule size distribution but in this study the two are decoupled. With increasing segregation, granule size distribution tightens at each bed depth, and this gives rise to a larger bed voidage. At present, the relationship between bed voidage and granule size distribution is still unknown for the sinter bed, and is assumed uniform in the segregated bed. The relationship between bed voidage, bulk density and granule apparent density is:²²⁾   

$${\psi }_{bulk}={\psi }_{apparent}\left(1-\epsilon \right)$$(4)

where, ψ_bulk is the bed bulk density, ψ_apparent the apparent density of granules and ε the bed voidage.

Bulk density decreases as bed voidage increases, assuming that granule apparent density is a constant. O’Dea and Waters⁸⁾ showed that increasing segregation only altered bed voidage slightly. When voidage increases from 0.383 to 0.394 – a 2.9% increase - bed bulk density decreased by 1.8%. However, this small change in bulk density can cause an increase in flame front speed.^1,23) In practice, increasing voidage will also increase sintering airflow rate, leading to increased flame front speed but this is not considered in this case. The effects of changing voidage are only considered in isolation.

Sensitivity analysis carried out on the developed iron ore sintering model has shown that a 10% increase in bed voidage can increase sintering airflow rate by 14%, which in turn can result in 12% increase in FFS. By assuming constant granule apparent density, this same 10% increase in voidage can decrease bulk density by 6%, leading to a 6% increase in FFS. The cumulative impact of these two factors indicates that every 10% increase in voidage can increase FFS by 18%. Overall, these considerations indicate that 10% decrease in bulk density (at constant granule apparent density) can give rise to 30% increase in flame front speed. Experimentally, Lovel et al.²⁴⁾ studied the effect of green bed bulk density on FFS in their sinter pot tests. As the bulk density was decreased from 2247 to 1752 kg m⁻³ by using different fuels, the FFS was found to increase greatly from 18 to 29 mm min⁻¹. Although fuel reactivity also influences FFS, their experimental results strongly suggest that green bed bulk density influences FFS.

In this study four uniform beds with the following voidages are considered: 0.383, 0.387, 0.390 and 0.394. Figure 7 shows the effect of voidage (and bulk density) on MT, η, FFS, EA and RT. All other parameters were set equal to base case (non-segregated) values in these simulations. It can be seen that MT, η and FFS all increased with decreasing bulk density, but both EA and RT dropped. The increased MT is a result of reduced volumetric heat capacity of the bed due to lower bed bulk density. The faster heating up of the bed also resulted in increased flame front speed. Higher flame front temperatures also resulted in the improvements in coke combustion efficiency η. The finding that EA and RT decreased with increasing bulk density is also consistent with the sinter pot test results,²³⁾ and is mainly the outcome of faster flame front speeds.

Fig. 7.

Effect of bed voidage segregation level σ_v on sintering performance indicators.

4.3. Effect of Coke Mass Segregation

Segregation causes coke levels to decrease down a bed. Based on the data in Tables 1 and 2, simulations with increasing coke mass segregation levels i.e., σ_m = 0, 7, 14 and 21% were carried out. To obtain a more extensive understanding of the effect of coke mass segregation on sintering performance, the range was further extrapolated to include the levels of 28, 35, 42 and 49%. As in Sections 4.1 and 4.2 all other parameters were set unchanged from the base case. Because coke has such a large impact on the sintering process, some results obtained at the three bed levels will also be presented. Figure 8 shows that MT, RT, EA and η values in the top and middle layers increase with increasing coke levels. As expected there are corresponding decreases in MT values in the bottom bed. The other changes in the lower bed temperature profile have meant that RT values increased slightly and EA values dropped, being more dependent on MT than RT values. Averaged values for the entire bed show that the three parameters increase in value with segregation. The overall combustion efficiency, η, also improves with segregation, which can lead to decreased fuel requirements.

Fig. 8.

Effect of coke mass segregation level σ_m on sintering performance indicators.

In Fig. 8(a) the MT graphs for the top and bottom bed crossed over at an overall segregation value of about 7%, where bed temperatures in the two regions are comparable at around 1340°C. It is obvious that very high segregation levels of, say, 50% is not desirable because of the excessively large difference in temperature between these two regions. Clearly the optimum segregation level based on the simulation should be around 7% and this is confirmed by the results in Fig. 9. MT values between the top and bottom thermocouples (Fig. 3) at different coke segregation levels σ_m are shown in Fig. 9(a). With increasing segregation, top bed temperature increases and bottom bed temperature decreases. Figure 9(b) shows the difference between the maximum and minimum MT values. There exists an optimum coke segregation level where the difference is close to zero. This segregation level is around 7% and the difference in MT value is 5°C. For 0% and 49% segregation levels, the differences are 24 and 82°C, respectively. For the non-segregated case, the temperature difference is mainly attributed to the preheating of incoming air by the hot sinter in the upper bed.¹⁾ This is a common phenomenon in the heat transfer in porous media.²⁵⁾ While for 49% segregation, the major factor causing the large temperature difference is the very high concentration of coke in a particular area of the bed.

Fig. 9.

Effect of coke segregation on (a) the distribution of MT between the top and bottom thermocouples, (b) difference between the maximum and minimum MT.

A heat balance over the entire sintering period has been carried out. Results indicate that coke mass segregation mainly changes the evolution of sensible heat and coke combustion heat, while other heats are only slightly altered due to the change of FFS and bed temperature. The optimum coke mass segregation level is achieved when the rate of sensible heat accumulated in the bed is not uniform. It is desirable to accumulate sensible heat at a higher rate in the earlier stage of sintering, and at a reduced rate in the later stage. This conclusion is consistent with the view that upper bed is the weakest region because of lower temperatures.

In this work, only coke mass segregation is considered. In practice, however, coke size segregation may also take place. In that case, large coke particles which are likely to be covered by adhering fine particles tend to accumulate in the lower bed.^2,26) This influence will be studied in future work.

4.4. Effect of Sintering Airflow Rate

The effect of increasing sintering airflow rate (or bed permeability) on sintering performance has been simulated. Results show that MT, η, EA and RT values decrease but FFS increased sharply with increasing sintering airflow rate. This indicates that increasing bed permeability has the potential to increase sintering productivity through decreasing sintering time.

4.5 Combined Effect

The combined effect of granule size, bed voidage and chemical segregation as well as increased sintering airflow rate can be simulated simply by combining all the changes in the parameters shown in Table 2. In doing this, it is assumed that each variable changes linearly with γ. The simulated results are shown in Fig. 10. It is seen that with increasing segregation levels MT, RT, EA, FFS and η all increased. These changes are advantages from a sinter yield and productivity perspective. This finding is also consistent with the pot tests and plant results.⁸⁾ It is possible to break down the results to determine the individual contributions of the four segregation parameters. For γ of 4 the four components are summarised in Table 6 and given the notation ‘combined parameters study’.

Fig. 10.

Combined effect of segregation on sintering performance indicators.

Table 6. Summary of effect of various parameters on sintering performances.

Parameter	MT (°C)	RT (%)	EA (%)	η (-)	FFS (mm min−1)
Combined parameters study	16.51	4.13	11.19	0.89	1.36
Single parameter study
Granule size	2.29	1.81	3.51	1.20	0.28
Bed voidage (and bulk density)	1.18	−1.81	−1.15	0.10	0.38
Coke mass	14.28	13.18	19.08	0.24	−0.77
Sintering airflow rate	−1.32	−7.75	−7.50	−0.37	1.52
Sum of above contributions	16.43	5.43	13.94	1.17	1.41

Note: Changes of RT and EA are shown on percentage basis, others are absolute values.

In sections 4.1 to 4.4, the four parameters affected by segregation were studied separately and results for γ of 4 are also shown in Table 6 under ‘single parameter study’. It is seen that coke mass segregation greatly raises the values of MT, RT and EA. Increasing sintering airflow rate has the second largest impact but lowers MT, RT and EA values. The effects of uneven distribution of granule size, increasing bed voidage and decreasing bulk density are comparatively small. Increasing all the segregation parameters except sintering airflow rate resulted in higher η. Results also show that increasing bed permeability and sintering airflow rate has a major effect on FFS. The other factor having some influence on FFS is coke segregation. The total contributions of all these influences can be added and results are also shown in Table 6 in the row ‘Sum of above contributions’. On comparing values in this row with the ‘Combined parameters study’ row it is seen that there are differences and this is not surprising. However, the relative contributions of the four segregation parameters to both results are not very different.

5. Conclusions

As granules are charged onto a moving sinter strand to form a bed, size and material type segregation occurs down the bed. This study investigates the effect of segregation on the properties of the critical melt formation region in and around the flame front using a well-validated mathematical model. The positive impact of segregation on MT, EA, RT and η indicates the potential to decrease fuel rate in sintering practice.

Specifically, vertical changes in bed properties caused by increased mean granule size are increased bed voidage and permeability. An outcome of these changes is a reduction in bed bulk density, which has been shown to increase the speed of the descending flame front. These factors together with changes in coke mass distribution will alter not only the amount of heat generated but also the combustion efficiency of the coke particles and properties of the critical melt formation region in the bed.

Results indicate that the individual effects of granule size and bed bulk density changes on these predicted parameters are minor. But the effect of improved bed permeability and coke mass segregation are more significant. As a result, faster flame front speed and increased maximum temperature, residence time and enclosed area were obtained with segregation. Higher combustion efficiency was also obtained, leading to the generation of more heat and decreased fuel rate.

The main effect of coke mass segregation is to change the evolution of sensible heat and coke combustion heat. The other heat components only change slightly because of variations in flame front speed and bed temperature. The accumulation of more sensible heat in the earlier rather than the later stages of sintering i.e., more heat generation in the upper bed is clearly very beneficial in terms of improving sintering performance. This study indicated that a coke segregation level of 7% - upper bed has more coke by this amount and the lower bed has less coke by this amount and the change occurs linearly down the bed - gave comparable bed temperature profiles at every level down the bed. However, the optimum coke segregation level may not be a constant value. This value may change with the factors that can influence the heat treatment process, for example, iron ore blend and bed height.

The results of this study provide important information on the effect of granule size, voidage and bulk density segregation on the properties of the flame front. The modelling studies also provide indicative coke segregation levels for more uniform heat treatment down a sintering bed.

Acknowledgements

The authors acknowledge the funding of the Australian Research Council in supporting the ARC Research Hub for Advanced Technologies for Australian Iron Ore and BHP Billiton for the financial support and permission to publish this paper.

References

• 1)   H.  Zhou,  J. P.  Zhao,  C. E.  Loo,  B. G.  Ellis and  K. F.  Cen: ISIJ Int., 52 (2012), 2168.
• 2)   C. E.  Loo: Ironmaking Steelmaking, 18 (1991), 33.
• 3)   D. F.  Ball,  J.  Dartnell,  J.  Davison,  A.  Grieve and  R.  Wild: Agglomeration of Iron Ores, Heinemann, London, (1973), 112.
• 4)   E.  Aries,  D. R.  Anderson,  R.  Fisher,  T. A. T.  Fray and  D.  Hemfrey: Chemosphere, 65 (2006), 1470.
• 5)   J. P.  Zhao,  C. E.  Loo and  R. D.  Dukino: Combust. Flame, 162 (2015), 1019.
• 6)   Y.  Arikata,  K.  Yamamoto and  Y.  Sassa: ISIJ Int., 53 (2013), 1523.
• 7)   H.  Zhou,  J. P.  Zhao,  C. E.  Loo,  B. G.  Ellis and  K. F.  Cen: ISIJ Int., 52 (2012), 1550.
• 8)   D.  O’Dea and  A.  Waters: Ironmaking Conf. Proc., ISS, Warrendale, PA, (1993), 459.
• 9)   D.  O’Dea and  A.  Waters: 4th Int. Conf. on Bulk Materials, Storage, Handling and Transportation: 7th Int. Symp. on Freight Pipelines, Institution of Engineers, Barton, (1992), 413.
• 10)   S.  Machida,  T.  Higuchi,  N.  Oyama,  H.  Sato,  K.  Takeda,  K.  Yamashita and  K.  Tamura: ISIJ Int., 49 (2009), 667.
• 11)   D.  Ball: Powder Technol., 9 (1974), 39.
• 12)   M.  Nakano,  T.  Abe,  J.  Kano and  K.  Kunitomo: ISIJ Int., 52 (2012), 1559.
• 13)   S.  Ishihara,  R.  Soda,  Q.  Zhang and  J.  Kano: ISIJ Int., 53 (2013), 1555.
• 14)   R.  Wakeman: Powder Technol., 11 (1975), 297.
• 15)   M.  Suzuki,  H.  Sato,  M.  Hasegawa and  M.  Hirota: Powder Technol., 118 (2001), 53.
• 16)   G.  Scott and  D.  Kilgour: J. Phys. D: Appl. Phys., 2 (1969), 863.
• 17)   H.  Masuda,  K.  Higashitani and  H.  Yoshida: Powder Technology: Fundamentals of Particles, Powder Beds, and Particle Generation, CRC Press, Florida, (2006), 293.
• 18)   M.  Fujimoto,  T.  Inazumi and  K.  Sato: Ironmaking Conf. Proc., ISS, Warrendale, PA, (1990), 589.
• 19)   C. E.  Loo and  M. F.  Hutchens: ISIJ Int., 43 (2003), 630.
• 20)   H.  Zhou,  J. P.  Zhao,  L. G.  Zheng,  C. L.  Wang and  K. F.  Cen: Eng. Appl. Artif. Intel., 25 (2012), 147.
• 21)   C. E.  Loo: ISIJ Int., 45 (2005), 436.
• 22)   J.  Hinkley,  A. G.  Waters,  D.  O’Dea and  J. D.  Litster: Int. J. Miner. Process., 41 (1994), 53.
• 23)   C. E.  Loo,  N.  Tame and  G. C.  Penny: ISIJ Int., 52 (2012), 967.
• 24)   R. R.  Lovel,  K. R.  Vining and M. Dell ‘Amico: ISIJ Int., 49 (2009), 195.
• 25)   S.  Zhdanok,  L. A.  Kennedy and  G.  Koester: Combust. Flame, 100 (1995), 221.
• 26)   C. S.  Teo,  R. A.  Mikka and  C. E.  Loo: ISIJ Int., 32 (1992), 1047.